报告时间：7月7日 星期一下午1:00
报告地点：浙江大学工控老楼414会议室
欢迎广大师生参加！
报告题目：间歇过程的多时段建模与过程监测
报告人：赵春晖
作为现代制造业中一种重要的生产方式，间歇工业过程与现代人的生活息息相关，被广泛应用于精细化工、生物制药、食品、聚合物反应、金属加工等领域。近年来，随着现代社会对多品种、多规格和高质量产品更迫切的市场需求，工业生产更加倚重于生产小批量、高附加值产品的间歇过程。因此，间歇生产的安全可靠运行以及产品的高质量追求已成为人们关注的焦点，而实施批次过程监测与质量分析则是实现该目标的重要手段。
与连续工业过程相比，间歇生产的过程特性更加复杂，同一操作批次内又分成多个子时段，每个时段都有其特定的控制目标，有不同的过程主导变量，呈现不同的过程相关特性。因此，面向多时段间歇生产过程的统计分析及在线应用更具挑战性，不仅仅要关注整个过程的运行状况及其与最终产品质量的因果关系，更应该深入分析过程的每一个子时段，揭示其不同的潜在过程特性并发掘它们对质量指标不同的影响效果和作用能力。本报告将在深入挖掘与研究间歇过程多时段特性的基础上，从解决实际问题的角度出发，结合注塑过程这一具体的实例，介绍如何捕捉过程动态性变化规律进而建立软划分与多时段建模理论方法以及基于多时段的间歇过程监测与质量分析方法。
报告题目：Asynchronous filtering for discrete-time stochastic Markov jump systems
报告人：吴争光
The problem of asynchronous energy-to-peak filtering is considered for discrete-time stochastic Markov jump systems with sensor nonlinearity. The sensor nonlinearity is assumed to occur randomly according to a stochastic variable satisfying the Bernoulli distribution. Two sufficient conditions are given such that the resultant filtering error system, which is a kind of nonhomogeneous Markov jump system, is stochastically stable with a guaranteed energy-to-peak performance index. Then the existence criterion of the desired asynchronous filter with piecewise homogeneous Markov chain is proposed in terms of a set of linear matrix inequalities. A numerical example is given to show the effectiveness and potential of the developed theoretical results.
报告题目：Resource Optimization in Networked Pervasive Sensing
报告人：贺诗波
Pervasive sensing is an essential component of cyber-physical system and is drawing great research attention. It is projected to double the existing measurement market. In this talk, we present two recent results on resource optimization in pervasive sensing. We first consider using camera sensor networks for security surveillance, and focus on full-view area coverage (a metric for quality of sensing) problem. We transform the full-view area coverage into full-view point coverage. After showing the formulated problem is NP hard, we propose an approximate algorithm with performance guarantee to solve the problem. Then we consider leverage the crowd with mobile users to do the sensing tasks. One challenge in the unified crowdsensing system is the task allocation problem. By including the unique requirement of crowdsensing, the task allocation turns out to be NP hard. A 5-approximate algorithm is proposed to solve the problem.
当前位置：